1.3

data/anchors.txt

@@ -1,11 +1,11 @@
-Best Accuracy = 77.39%
+Best Accuracy = 77.86%
-
+
-Best Anchors = [[ 87  51]
+Best Anchors = [[109  52]
- [ 78  85]
+ [ 78  81]
- [136  65]
+ [ 96 152]
- [ 99 152]
+ [139 106]
- [147 108]
+ [230  70]
- [238  72]
+ [160 172]
- [240 127]
+ [241 126]
- [173 180]
+ [217 202]
- [264 203]]
+ [307 201]]

detect.py

@@ -150,7 +150,8 @@ def detect(opt, save_img=False):
              "6923644272159", "6924882486100", "6956511907458"]
     targets = []
     for target in pred[0]:
-        targets.append({"Class": names[int(target[5].item())], "precision": target[4].item(), "xy1": [target[0].item(), target[1].item()],
+        targets.append({"Class": names[int(target[5].item())], "precision": target[4].item(),
+                        "xy1": [target[0].item(), target[1].item()],
                         "xy2": [target[2].item(), target[3].item()]})
     resu = {"TargetDetect": targets}
     print(resu)

getval_imgs.py

@@ -1,7 +1,7 @@
 import os,shutil
 
 allimgs_path = "paper_data/images"
-valimgs_path = "paper_data/val_imgs"
+valimgs_path = "paper_data/val_imgs_7.1"
 valtxt_path = "paper_data/ImageSets/Main/val.txt"
 
 with open(valtxt_path, "r", encoding='UTF-8') as val_file:

ieemoo-ai-zhanting.py

@@ -40,7 +40,7 @@ parser = argparse.ArgumentParser()
 parser.add_argument('--weights', nargs='+', type=str, default='../module/ieemoo-ai-zhanting/model/now/best.pt', help='model.pt path(s)')
 parser.add_argument('--source', type=str, default='../module/ieemoo-ai-zhanting/imgs/1.jpg', help='source')  # file/folder, 0 for webcam
 parser.add_argument('--img-size', type=int, default=640, help='inference size (pixels)')
-parser.add_argument('--conf-thres', type=float, default=0.55, help='object confidence threshold')
+parser.add_argument('--conf-thres', type=float, default=0.60, help='object confidence threshold')
 parser.add_argument('--iou-thres', type=float, default=0.45, help='IOU threshold for NMS')
 parser.add_argument('--device', default='', help='cuda device, i.e. 0 or 0,1,2,3 or cpu')
 parser.add_argument('--view-img', type=bool, default=True, help='display results')

label_num.py

@@ -0,0 +1,30 @@
+###  计算xml中每个类别标签的数量  ###
+
+import os
+import shutil
+import xml.etree.ElementTree as ET
+
+def caculate_label_numble(file_path):
+    label_numble_info = {}
+    for file in os.listdir(file_path):
+        if file.endswith(".xml"):
+            xml_path = os.path.join(file_path, file)
+            tree = ET.parse(xml_path)
+            root = tree.getroot()
+
+            for element in root.findall('object'):
+                for element1 in element.findall('name'):
+                    if element1.text not in label_numble_info:
+                        # if element1.text=="6935284412918":#查看异常分类对应的xml
+                        #     print(file)
+                        label_numble_info[element1.text] = 1
+                    else:
+                        label_numble_info[element1.text] += 1
+    return label_numble_info
+
+if __name__=="__main__":
+
+    file_path = "paper_data/Annotations"
+    label_numble_info = caculate_label_numble(file_path)
+    for key, value in label_numble_info.items():
+        print('{}:{}'.format(key, value))

parseXml.py

@@ -0,0 +1,191 @@
+# -*- coding: utf-8  -*-
+# -*- author: jokker -*-
+import os
+import time
+import numpy as np
+import xml.etree.ElementTree as ET
+from xml.dom.minidom import parseString
+from lxml.etree import Element, SubElement, tostring
+from xml.etree.ElementTree import fromstring, ElementTree
+from xml.etree import ElementTree
+
+cwd=os.getcwd()
+def prettyXml(element, indent, newline, level=0):  # elemnt为传进来的Elment类，参数indent用于缩进，newline用于换行
+    if element:  # 判断element是否有子元素
+        if element.text == None or element.text.isspace():  # 如果element的text没有内容
+            element.text = newline + indent * (level + 1)
+        else:
+            element.text = newline + indent * (level + 1) + element.text.strip() + newline + indent * (level + 1)
+            # else:  # 此处两行如果把注释去掉，Element的text也会另起一行
+    # element.text = newline + indent * (level + 1) + element.text.strip() + newline + indent * level
+    temp = list(element)  # 将elemnt转成list
+    for subelement in temp:
+        if temp.index(subelement) < (len(temp) - 1):  # 如果不是list的最后一个元素，说明下一个行是同级别元素的起始，缩进应一致
+            subelement.tail = newline + indent * (level + 1)
+        else:  # 如果是list的最后一个元素， 说明下一行是母元素的结束，缩进应该少一个
+            subelement.tail = newline + indent * level
+        prettyXml(subelement, indent, newline, level=level + 1)  # 对子元素进行递归操作
+    return element
+
+
+
+class ParseXml(object):
+    """解析 xml 中的信息，将信息导出为 xml"""
+
+    def __init__(self, input_xml):
+        self.__ttrs = {"folder", "filename", "path", "segmented", "size", "source", "object"}  # 所有的属性
+        self.__xml_info_dict = {}  # xml 信息字典
+        # self.__objects_info = []
+        # self.__size_info = {}
+        # self.__source_info = {}
+        self.input_xml = input_xml
+
+    def get_info(self):
+        tree = ET.parse(self.input_xml)
+        root = tree.getroot()
+        for name in self.__ttrs:
+            for element in root.findall(name):
+                print(element.text)
+
+    def get_xml_info(self): #获取xml中的信息，以字典形式保存
+        """获取xml中所有信息，并以字典形式呈现"""
+        xml_info = {}
+        tree = ET.parse(self.input_xml)
+        root = tree.getroot()
+
+        for element in root.findall('folder'):
+            xml_info["folder"]=element.text
+        for element in root.findall('filename'):
+            xml_info['filename'] = element.text
+        for element in root.findall('path'):
+            xml_info['path'] = element.text
+
+        for element in root.findall('source'):
+            for element1 in element.findall('database'):
+                xml_info['database'] = "Unknown"
+
+        for element in root.findall('size'):
+            for element1 in element.findall('width'):
+                xml_info['width'] = element1.text
+            for element1 in element.findall('height'):
+                xml_info['height'] = element1.text
+            for element1 in element.findall('depth'):
+                xml_info['depth'] = element1.text
+
+        for element in root.findall('segmented'):
+            xml_info['segmented'] = 0
+        boxes=[]
+        for element in root.findall('object'):
+            box={}
+            for element1 in element.findall('name'):
+                box['name'] = element1.text
+            for element1 in element.findall('pose'):
+                box['pose'] = element1.text
+            for element1 in element.findall('truncated'):
+                box['truncated'] = 0
+            for element1 in element.findall('difficult'):
+                box['difficult'] = 0
+            for element1 in element.findall("bndbox"):
+                box['bndbox']={}
+                for element2 in element1.findall("xmin"):
+                    box['bndbox']['xmin'] = element2.text
+                for element2 in element1.findall("ymin"):
+                    box['bndbox']['ymin'] = element2.text
+                for element2 in element1.findall("xmax"):
+                    box['bndbox']['xmax'] = element2.text
+                for element2 in element1.findall("ymax"):
+                    box['bndbox']['ymax'] = element2.text
+            boxes.append(box)
+        xml_info['object']=boxes
+        return xml_info
+
+    def write_to_xml(self,result,save_dir):
+        """
+        将目标信息写入到xml中
+        """
+        # if not os.path.exists(save_dir):
+        #     os.mkdir(save_dir)
+        imgname=os.path.splitext(result['filename'])[0]
+        xmlPath = os.path.join(save_dir, imgname)+".xml"
+
+        root = ET.Element("annotations")
+        ET.SubElement(root, "folder").text = result['folder']
+        ET.SubElement(root, "filename").text = result['filename']
+        ET.SubElement(root, "path").text = result['path']
+        size = ET.SubElement(root, "size")
+        ET.SubElement(size, "width").text = str(result['width'])
+        ET.SubElement(size, "height").text = str(result['height'])
+        ET.SubElement(size, "depth").text = "3"
+
+        for info in result['object']:
+            obj = ET.SubElement(root, "object")
+            ET.SubElement(obj, "name").text = str(info['name'])
+            ET.SubElement(obj, "pose").text = str(info['pose'])
+            ET.SubElement(obj, "truncated").text = str(info['truncated'])
+            ET.SubElement(obj, "difficult").text = str(info['difficult'])
+
+            bbox = ET.SubElement(obj, "bndbox")
+            ET.SubElement(bbox, "xmin").text = str(int(float(info['bndbox']['xmin'])))
+            ET.SubElement(bbox, "ymin").text = str(int(float(info['bndbox']['ymin'])))
+            ET.SubElement(bbox, "xmax").text = str(int(float(info['bndbox']['xmax'])))
+            ET.SubElement(bbox, "ymax").text = str(int(float(info['bndbox']['ymax'])))
+
+        tree = ET.ElementTree(root)
+        tree.write(xmlPath)
+
+        tree = ElementTree.parse(xmlPath)
+        root = tree.getroot()  # 得到根元素，Element类
+        root = prettyXml(root, '\t', '\n')  # 执行美化方法
+        # ElementTree.dump(root)  # 打印美化后的结果
+        tree = ET.ElementTree(root)  # 转换为可保存的结构
+        tree.write(xmlPath)  # 保存美化后的结果
+
+        return xmlPath
+
+    def update_node(self,xml,save_dir,note,note_value):
+        """
+        更改xml中的任意节点，并保存新的xml
+        """
+        # if not os.path.exists(save_dir):
+        #     os.mkdir(save_dir)
+        imgname = os.path.basename(xml)
+        xmlPath = os.path.join(save_dir, imgname)
+
+        tree = ElementTree.parse(xml)
+        root = tree.getroot()  # 得到根元素，Element类
+        try:
+            if note in ["folder", "filename", "path"]:
+                for element in root.findall(note):
+                    element.text = note_value
+            elif note in ["width", "height", "depth"]:
+                for element in root.findall("size"):
+                    for element1 in element.findall(note):
+                        element1.text = str(note_value)
+            elif note in ["name", "pose", "truncated", "difficult"]:
+                for element in root.findall("object"):
+                    for element1 in element.findall(note):
+                        element1.text = str(note_value)
+            elif note in ["xmin", "xmax", "ymin", "ymax"]:
+                for element in root.findall("object"):
+                    for element1 in element.findall("bndbox"):
+                        for element2 in element1.findall(note):
+                            element2.text = str(note_value)
+        except Exception as error:
+            print("修改的节点不在xml")
+        tree.write(xmlPath)
+#
+
+
+
+if __name__ == "__main__":
+
+    xml_path = 'C:\\Users\\Administrator\\Desktop\\1.xml'
+    save_dir = 'C:\\Users\\Administrator\\Desktop\\merge1'
+    xmlInfo = ParseXml(xml_path)
+    result = xmlInfo.get_xml_info()
+    print(result)
+
+    xmlPath = xmlInfo.write_to_xml(result,save_dir)
+    print("--*--xmlPath--*--:", xmlPath)
+    xmlInfo.update_node(xmlPath, save_dir, "xmin", 200)
+

spilt_train_val.py

@@ -30,14 +30,22 @@ file_test = open(txtsavepath + '/test.txt', 'w')
 file_train = open(txtsavepath + '/train.txt', 'w')
 file_val = open(txtsavepath + '/val.txt', 'w')
 
+addtrain_path = r"D:\PycharmProjects\Zhanting\yolov5_1\img_data\getimg_6.30"
 for i in list_index:
     name = total_xml[i][:-4] + '\n'
+    addimg_name = name.strip() + ".jpg"
+    # print(addimg_name,type(addimg_name),len(addimg_name))
     if i in trainval:
         file_trainval.write(name)
         if i in train:
             file_train.write(name)
         else:
-            file_val.write(name)
+            if addimg_name in os.listdir(addtrain_path):#把某些被划分到验证集的数据加入训练集中
+                print("addimg_name:",addimg_name)
+                file_train.write(name)
+            else:
+                file_val.write(name)
+            # file_val.write(name)
     else:
         file_test.write(name)